Adaptive compensation lane keeping system and method

Abstract

This invention provides an adaptive compensation lane keeping system and method, belonging to the field of intelligent driving. The system includes: an input module, a calculation module, a judgment module, and an output module connected in sequence; the input module is used to input parameters; the calculation module is used to calculate the angle compensation base value, the torque offset compensation base value, the LKA requested angle average value, the C0 average value, and the attenuation coefficient; the judgment module is used to determine whether compensation should be performed based on the LKA requested angle average value and the C0 average value; if so, it determines the compensation direction, adds the compensation direction before the angle compensation base value to obtain the angle compensation value, and adds the compensation direction before the torque offset compensation base value to obtain the torque offset compensation value; the output module is used to obtain the adaptively compensated planned angle based on the angle compensation value and the attenuation coefficient, and sends the adaptively compensated planned angle and the torque offset compensation value to the EPS. This invention achieves lane centering control.

Description

Technical Field

[0001] This invention belongs to the field of intelligent driving, specifically relating to an adaptive compensation lane keeping system and method. Background Technology

[0002] Currently, vehicles equipped with driver assistance cameras can implement Lane Keeping Assist (LKA) by controlling the Electronic Power Steering (EPS) system to provide steering control for the driver and assist the driver in keeping the vehicle within its lane. The main benefit for the driver is reduced steering load and improved driving comfort.

[0003] Torque request control is a common LKA request control method. The main logic is as follows: the EPS receives the torque applied to the steering wheel by the driver and the torque requested by the LKA function from the motor. The EPS then processes these signals to obtain the total torque applied to the motor. The accuracy of the LKA controller's control signal (steering torque) calculation directly determines the control performance.

[0004] Torque-controlled LKA systems are susceptible to interference from EPS zero-point torque (due to manufacturing inconsistencies causing the steering wheel motor torque to be non-zero when the steering wheel returns to center from a stationary position). Excessive zero-point torque can cause the vehicle to swerve during LKA control, affecting centering accuracy. Simultaneously, road bumps can easily alter the steering wheel motor torque under torque-controlled LKA systems, leading to swerving, off-center driving, and excessive adjustments. Furthermore, after a certain mileage (e.g., exceeding 50,000 km) or frequent driving on mountain roads, the chassis can become loose, leading to inaccurate four-wheel alignment parameters. A loose chassis can cause the vehicle to veer off course, reducing control precision and resulting in off-center driving.

[0005] Angle control is another LKA (Lane Assist Key) request control method that has been increasingly adopted by OEMs recently. Its main logic is as follows: the EPS (Electric Power Steering) receives the torque value applied to the steering wheel by the driver and the angle request value requested by the LKA function on the motor. This is then converted into steering torque to control the EPS motor via the EPS's gear angle controller. However, angle control suffers from discomfort when the driver takes over. The steering feel is stiff and abrupt, making it difficult for the driver to intervene, or it may be resistant, requiring considerable force to take over. Furthermore, after the vehicle has traveled a certain distance, the chassis may loosen, causing the vehicle to veer off course, and the angle control accuracy will decrease, potentially leading to the vehicle not staying centered. Summary of the Invention

[0006] The purpose of this invention is to solve the problems existing in the prior art and provide an adaptive compensation lane keeping system and method, which solves the problems of torque control being disturbed by zero torque and road bumps, and solves the problem of LKA control accuracy decreasing due to vehicle deviation, as well as the problems of driver takeover feeling hard, abrupt, unable to intervene, or resistance to takeover requiring a lot of force.

[0007] This invention is achieved through the following technical solution:

[0008] In a first aspect, the present invention provides an adaptive compensation lane keeping system, the system comprising: an input module, a calculation module, a judgment module and an output module connected in sequence;

[0009] The input module is used to input parameters;

[0010] The calculation module is used to calculate the angle compensation base value, torque offset compensation base value, LKA requested angle mean, C0 mean, and attenuation coefficient;

[0011] The judgment module is used to determine whether to perform compensation based on the average angle value and average C0 value requested by LKA. If so, the compensation direction is determined, and the compensation direction is added before the angle compensation base value to obtain the angle compensation value. The compensation direction is added before the torque offset compensation base value to obtain the torque offset compensation value.

[0012] The output module is used to obtain the adaptively compensated planned angle based on the angle compensation value and the attenuation coefficient, and send the adaptively compensated planned angle and the torque offset compensation value to the EPS.

[0013] Preferably, the parameters input to the input module include:

[0014] Vehicle speed V, yaw rate, relative distance C0 between the estimated driving line of the current vehicle position and the vehicle body, heading angle C1, curvature of the target trajectory at the current point of the vehicle C2, rate of change of curvature of the target trajectory C3, steering wheel torque, LKA control status, steering wheel angle, LKA request angle.

[0015] A second aspect of the present invention provides an adaptive compensation lane keeping method, the method comprising:

[0016] Step 1, input parameters: vehicle speed V, yaw rate, relative distance C0 between the estimated driving line of the current vehicle position and the vehicle body, heading angle C1, curvature of the target trajectory at the current point of the vehicle C2, rate of change of curvature of the target trajectory C3, steering wheel torque, LKA control status, steering wheel angle, LKA request angle;

[0017] The second step is to calculate the base value of angle compensation, the base value of torque offset compensation, the average value of LKA requested angle, the average value of C0, and the attenuation coefficient.

[0018] The third step is to determine whether compensation is required. If so, determine the compensation direction, then add the compensation direction before the angle compensation base value to obtain the angle compensation value, and add the compensation direction before the torque offset compensation base value to obtain the torque offset compensation value, and then proceed to the fourth step; if not, return to the second step.

[0019] The fourth step is to obtain the adaptively compensated planned angle based on the angle compensation value and the attenuation coefficient, and then send the adaptively compensated planned angle and the torque offset compensation value to the EPS.

[0020] In the fifth step, EPS converts the adaptively compensated planned angle into planned torque, then adds the planned torque to the torque offset compensation value to obtain the total executed torque, and then returns to the second step.

[0021] A further improvement of the present invention is that:

[0022] The second step involves calculating the baseline values ​​for angle compensation and torque offset compensation, including:

[0023] (A1) Settings: Calculation count x = 0, Monitoring count y = 0, Memory value = 0;

[0024] (A2) Determine whether the condition is met. If yes, proceed to step (A3); otherwise, proceed to step (A13).

[0025] (A3) Calculate the mean over 30 seconds and assign it to N(x);

[0026] (A4) x = x + 1;

[0027] (A5) Determine whether x < the first threshold is true. If yes, return to step (A2). If no, proceed to step (A6).

[0028] (A6) Determine whether x = 1 is true. If yes, proceed to step (A7); otherwise, proceed to step (A11).

[0029] (A7) Mean = N(x-1);

[0030] (A8) Determine whether the absolute value of the difference between the mean and the memory value is greater than the set threshold. If yes, proceed to step (A9); otherwise, proceed to step (A10).

[0031] (A9) Compensation base value = mean, memory value = compensation base value; then proceed to step (A15);

[0032] (A10) Compensation base value = 0, memory value = 0, then proceed to step (A15);

[0033] (A11) Determine whether x = 0 is true. If it is, return to step (A10). If not, proceed to step (A12).

[0034] (A12) Mean = 0.5N(x-1) + 0.5N(x-2), then return to step (A8);

[0035] (A13), y = y + 1;

[0036] (A14) Determine whether y < the second threshold is true. If yes, return to step (A2); otherwise, return to step (A6).

[0037] (A15) End.

[0038] A further improvement of the present invention is that:

[0039] The second step, calculating the mean LKA request angle and the mean C0, includes:

[0040] (B1) Settings: Calculation count x = 0, Monitoring count y = 0;

[0041] (B2) Determine whether the condition is met. If yes, proceed to step (B3); otherwise, proceed to step (B11).

[0042] (B3) Calculate the mean over 30 seconds and assign it to N(x);

[0043] (B4) x = x + 1;

[0044] (B5) Determine whether x < the first threshold is true. If yes, return to step (B2). If no, proceed to step (B6).

[0045] (B6) Determine whether x = 1 is true. If yes, proceed to step (B7); otherwise, proceed to step (B8).

[0046] (B7) Mean = N(x-1), then proceed to step (B13);

[0047] (B8) Determine whether x = 0 is true. If yes, proceed to step (B9); otherwise, proceed to step (B10).

[0048] (B9) Mean = 0, then proceed to step (B13);

[0049] (B10) Mean = 0.5N(x-1) + 0.5N(x-2), then proceed to step (B13);

[0050] (B11), y = y + 1;

[0051] (B12) Determine whether y < the second threshold is true. If yes, return to step (B2); otherwise, return to step (B6).

[0052] (B13) End.

[0053] A further improvement of the present invention is that:

[0054] The operations for determining whether the conditions are met in steps (A3) and (B3) include:

[0055] If all four of the following conditions are met simultaneously and last for 30 seconds, the conditions are considered met; otherwise, the conditions are considered not met:

[0056] Condition 1: |C1| < 0.02;

[0057] Condition 2: |C2| < 0.00003;

[0058] Condition 3: V > 10 kph;

[0059] Condition 4: A 1-second delay after the steering wheel torque is less than 0.5 Nm.

[0060] A further improvement of the present invention is that:

[0061] The second step of calculating the attenuation coefficient includes:

[0062] The attenuation coefficient α is calculated using the following formula:

[0063] α = (1-β) × 100%

[0064] β=1, when T C ≤T min hour

[0065] When T min <T C <T max hour

[0066] β=0, when T C ≥T max hour

[0067] Among them, T c T represents the current steering wheel torque. min T is the lower limit of steering wheel torque. max β represents the upper limit of steering wheel torque, and β is the torque process variable.

[0068] A further improvement of the present invention is that:

[0069] The operation of determining whether to perform compensation in the third step includes:

[0070] (S1) Determine whether the average value of the LKA requested angle and the average value of C0 are both equal to 0. If yes, proceed to step (S3); otherwise, proceed to step (S2).

[0071] (S2) Determine whether the directions of the LKA requested angle mean and the C0 mean are opposite. If yes, determine to perform compensation and then proceed to step (S4). If no, proceed to step (S3).

[0072] (S3) It is determined that no compensation will be provided;

[0073] (S4), End.

[0074] A further improvement of the present invention is that:

[0075] The operation of determining the compensation direction in the third step includes:

[0076] Determine if the mean of C0 is greater than the threshold. If it is, the compensation direction is determined to be negative; otherwise, the compensation direction is determined to be positive.

[0077] A further improvement of the present invention is that:

[0078] The operation in the fourth step of obtaining the adaptively compensated planning angle based on the angle compensation value and the attenuation coefficient includes:

[0079] The adaptively compensated planning angle θt is calculated using the following formula:

[0080] θt=θc+α(θp-θc)K+θa,

[0081] Where: θc is the current steering wheel angle, θp is the planned angle, θa is the angle compensation value, and K = calibration constant coefficient.

[0082] Compared with the prior art, the beneficial effects of the present invention are: the present invention replaces the original torque control with angle control, realizes lane centering control, avoids the problems of torque control being affected by zero torque and road surface bumps; and improves the stability of driving in the center.

[0083] This invention avoids the problem of decreased LKA control accuracy caused by vehicle deviation;

[0084] This invention avoids the problems of driver taking over the steering wheel feeling stiff, abrupt, unable to intervene, or resistance to taking over the steering wheel, requiring considerable force to do so. Attached Figure Description

[0085] Figure 1 Block diagram of the LKA working system with angle control;

[0086] Figure 2 For information processing and decision-making control processes;

[0087] Figure 3 A structural diagram of the adaptive compensation lane keeping system of this invention;

[0088] Figure 4-1 The process for calculating the base value of compensation;

[0089] Figure 4-2 The process for calculating the mean of C0 and the mean of LKA request angles;

[0090] Figure 5-1 The process for determining whether compensation should be provided;

[0091] Figure 5-2 The process for determining the direction of compensation;

[0092] Figure 6 Schematic diagram of angle control error. Detailed Implementation

[0093] The present invention will now be described in further detail with reference to the accompanying drawings:

[0094] Currently, the existing LKA system architecture for angle control methods is as follows: Figure 1 As shown, it comprises a perception layer, a processing layer, a control layer, and an execution layer. The perception layer uses a lane departure detection system based on a video imaging system to measure the distance and orientation of the vehicle relative to the left and right lane lines, and outputs lane line information, vehicle status information (including steering angle, yaw rate, driver's hand torque, etc.), EPS status, and driver status information to the processing layer. The processing layer derives the vehicle's positional relationship with the lane lines and steering wheel status information based on the lane line information and vehicle information. The control layer is the LKA controller, which determines the LKA status and requested torque angle based on the vehicle's positional relationship with the lane lines and steering wheel status information input from the processing layer. The LKA execution layer consists of the EPS control system and the instrument display system.

[0095] The LKA control algorithm is briefly described below:

[0096] Path tracking control is a key technology for lane keeping control. The model predicts and plans the path for the next stage in advance, and makes assessments, decisions and executes in advance. It can overcome the impact of model errors and uncertain environmental interference, and has strong robustness and good real-time control.

[0097] Figure 2 for Figure 1The workflow of the processing and control layers is as follows: Road information is acquired through driver assistance cameras. The cameras receive vehicle status information, such as steering wheel angle and torque provided by EPS, vehicle speed and yaw rate provided by ESP, and lane line position information such as curvature and width, as perceived and output by the cameras, converted to C0, C1, C2, and C3. Using C0, C1, C2, C3, vehicle speed, steering wheel angle, and steering wheel torque information, the cameras combine this with the vehicle's driving trajectory to perform path tracking control. The feedforward controller calculates the feedforward angle compensation, which is then input to the feedback controller. The feedback controller, combined with the angle request value, can plan the path for the next stage of driving in advance; this results in the first closed-loop control of the driving system (e.g., ...). Figure 2 The "+" sign between the "feedforward controller" and the "feedback controller" in the diagram indicates the second closed-loop control (as shown in the diagram). Figure 2 (As shown by the "+" sign on the right side of "Speed ​​Control Loop PI Control"). Adding feedforward control to the speed closed-loop control effectively reduces control lag. Position loop P control calculates the difference between the target steering angle and vehicle speed using a lookup table. It employs a fuzzy algorithm based on error and vehicle speed, separating large-angle and small-angle request conditions, effectively addressing the request and response issues in large-angle situations. Speed ​​loop PI control uses a lookup table based on the deviation between the target and actual steering angles, employing integral I parameters (in process control, the PID controller, which controls based on the proportional (P), integral (I), and derivative (D) of the deviation, is the most widely used automatic controller). This solves the problem of P control being highly susceptible to road surface interference, especially on steep curves or slopes, where simply using P control results in large steering wheel rotation, slow convergence, and uncomfortable control. Furthermore, to avoid overshoot due to excessively large integral I parameters, integral I trigger conditions and clearing operations are added. The feedforward controller calculates the target turning angle using a lookup table. By adding feedforward control, the strength of the feedback control is weakened, effectively reducing interference from the control during cornering and on inclined roads. This forms a third closed-loop control, such as... Figure 2 The "+" signs to the right of "Angle Planning Control" and "Position Loop P Control" are shown in the diagram. The feedforward angle compensation output from the feedforward controller is superimposed with the LKA angle request value output from the third closed-loop control to obtain the planned angle θp. Then, the planned angle θp is input to the EPS. The gear tooth angle controller of the EPS converts the planned angle θp into torque, which is used as the total torque of the EPS motor.

[0098] This invention incorporates adaptive compensation (i.e., ...) into the existing processing flow. Figure 2 (Autofix compensation in the code). Specifically, the planned angle θp is superimposed with the adaptive compensation and then input into the EPS to form a fourth control closed loop, such as... Figure 2The "+" sign to the right of "autofix compensation" is shown in the image.

[0099] Adaptive compensation can address the impact of vehicle deviation or zero-point torque outside the design threshold on LKA control stability. It mainly obtains a force that matches the vehicle's motion state, which can improve LKA control performance to a certain extent.

[0100] The adaptive compensation steps provided by this invention are as follows: Figure 3 As shown, adaptive compensation enables angle closed-loop control in conjunction with the EPS controller. Compared to other LKA controls, the adaptive compensation added in this invention enables an additional type of angle closed-loop control, resulting in better control accuracy, stability, and comfort.

[0101] Specifically, such as Figure 3 As shown, the adaptive compensation lane keeping system provided by this invention includes: an input module, a calculation module, a judgment module, and an output module connected in sequence. The input module is used to input parameters. The input parameters all come from data collected in real-time by the existing vehicle (the functions for collecting this data are already present in the existing vehicle, briefly described below: the camera receives vehicle status information, such as steering wheel angle and motor torque information provided by EPS, vehicle speed V and yaw rate provided by ESP, and C0, C1, C2, and C3 calculated by the camera using the lane line curvature and width output by perception). The input parameters include: vehicle speed V, yaw rate, C0, C1, C2, C3, steering wheel torque (driver's hand force), LKA control status, steering wheel angle, and LKA request angle. C0, C1, C2, and C3 are four parameters of a polynomial curve. C0 is the relative distance between the estimated driving line of the current vehicle position and the vehicle body (calculated by the camera based on the planned trajectory and half the width of the identified lane line), C1 is the heading angle θ... C1 is the angle between the camera's driving trajectory and the lane line, C2 is the curvature of the target trajectory at the current point of the vehicle (the curvature of the target trajectory at the current point of the vehicle is directly output by the camera), and C3 is the rate of change of curvature of the target trajectory (obtained by time difference based on C2). The data collected by the vehicle in real time is input into the input module in real time, and the input module sends the input parameters to the calculation module in real time.

[0102] The calculation module is used to calculate the baseline values ​​for angle compensation, torque offset compensation, LKA requested angle, C0, and attenuation coefficient. Specifically, the calculation module calculates the steering wheel angle and steering wheel torque (driver's hand force) sent by the input module using... Figure 4-1 The procedures shown obtain the baseline values ​​for angle compensation and torque offset compensation, respectively. Figure 4-1In this context, "angle compensation base value" and "torque offset compensation base value" are collectively referred to as compensation base value; the calculation module calculates the compensation base value based on the C0 and LKA angle requests sent by the input module. Figure 4-2 The processes shown obtain the mean C0 value and the mean LKA request angle value (in...) Figure 4-2 In this context, "C0 mean" and "LKA requested angle mean" are collectively referred to as the mean. The calculation module calculates the angle compensation base value, torque offset compensation base value, LKA requested angle mean, C0 mean, and attenuation coefficient in real time based on the parameters received in real time, and sends these data to the judgment module in real time.

[0103] The judgment module determines whether compensation should be performed based on the average angle and C0 value requested by LKA. If so, it determines the compensation direction (positive or negative), adds the compensation direction before the base angle compensation value to obtain the angle compensation value, and adds the compensation direction before the base torque offset compensation value to obtain the torque offset compensation value. Then, it sends the angle compensation value, torque offset compensation value, and attenuation coefficient to the output module. If not, it does not send any data to the output module. Specifically, the judgment module uses the process shown in Figure 5 for judgment.

[0104] The output module is used to obtain the adaptively compensated planned angle based on the angle compensation value and the attenuation coefficient, and send the adaptively compensated planned angle and the torque offset compensation value to the EPS.

[0105] The gear tooth angle controller in EPS uses an existing algorithm to convert the adaptively compensated planned angle into a planned torque. Then, the planned torque is superimposed with the torque offset compensation value to obtain the total execution torque, which is the total torque executed by the EPS motor.

[0106] The present invention also provides an adaptive compensation lane keeping method, the method comprising:

[0107] The first step is to input parameters. The input parameters include: vehicle speed V, yaw rate, C0, C1, C2, C3, steering wheel torque, LKA control status, steering wheel angle, and LKA request angle. These parameters are collected and input in real time.

[0108] The second step is to calculate the baseline value of angle compensation, the baseline value of torque offset compensation, the average value of LKA requested angle, the average value of C0, and the attenuation coefficient; the second step is to perform real-time calculations based on the parameters received in real time.

[0109] Specifically, such as Figure 4-1 As shown, the second step involves calculating the baseline values ​​for angle compensation and torque offset compensation, including:

[0110] (A1) Settings: Calculation count x = 0, Monitoring count y = 0, Memory value = 0;

[0111] (A2) Determine whether the condition is met. If yes, proceed to step (A3); otherwise, proceed to step (A13).

[0112] (A3) Calculate the mean over 30 seconds and assign it to N(x); the vehicle itself samples at regular intervals (e.g., 0.1 seconds). The mean over 30 seconds is obtained by summing the sampled data over 30 seconds and dividing by the number of samples. For example, to find the mean of the steering wheel angle, sum all the steering wheel angles sampled over 30 seconds and divide by the number of samples. Similarly, the mean of the zero-point torque of the steering wheel can be calculated. For the sake of simplicity, in... Figure 4-1 The text only states "the average value over 30 seconds".

[0113] (A4) x = x + 1;

[0114] (A5) Determine if x < the first threshold. If yes, return to step (A2); otherwise, proceed to step (A6). The first threshold is set according to actual needs. Based on practical experience, when the number of calculations approaches 10, the output compensation base value tends to stabilize. Therefore, to avoid continuously calculating without being able to output a specific compensation base value, Figure 4-1 In the embodiment shown, the first threshold is set to 10.

[0115] (A6) Determine whether x = 1 is true. If yes, proceed to step (A7); otherwise, proceed to step (A11).

[0116] (A7) Mean = N(x-1); Step (A7) indicates that only one calculation has been performed, that is, only N(0). At this time, N(0) can be assigned to the mean. Here, the "mean" refers to the mean of the steering wheel angle and the mean of the zero-position torque of the steering wheel.

[0117] (A8) Determine if the absolute value of the difference between the mean and the memory value is greater than the set threshold. If yes, proceed to step (A9); otherwise, proceed to step (A10). For example, for the mean steering wheel angle, the set threshold is 0.3 degrees. This is because experience shows that when the vehicle is not veering off course and is driving on a straight road, the zero-position deviation of the steering wheel angle can be within ±0.3 degrees. If the absolute value of the difference between the mean calculated in this round and the compensation base value obtained in the previous round is less than 0.3 degrees, no compensation is performed; otherwise, compensation is performed.

[0118] (A9) Compensation base value = mean value, memory value = compensation base value; then proceed to step (A15); in step (A9), “compensation base value” refers to angle compensation base value and torque offset compensation base value; “memory value” refers to angle compensation memory value and torque offset compensation memory value, which are the angle compensation base value and torque offset compensation base value obtained in the previous round of calculation.

[0119] (A10) Compensation base value = 0, memory value = 0, then proceed to step (A15); Entering step (A10) indicates that the absolute value of the difference between the mean and the compensation base value obtained in the previous round of calculation is less than or equal to the set threshold, or no calculation has been performed, that is, no supplementation is required. Therefore, in step (A10), the angle compensation base value and torque offset compensation base value are set to 0.

[0120] (A11) Determine whether x = 0 is true. If it is, return to step (A10). If not, proceed to step (A12).

[0121] (A12) Mean = 0.5N(x-1) + 0.5N(x-2), then return to step (A8); entering step (A12) indicates that more than one calculation has been performed. In step (A12), the mean values ​​obtained from the last two calculations are averaged and assigned to the mean value.

[0122] (A13), y = y + 1;

[0123] (A14) Determine if y < the second threshold is true. If yes, return to step (A2); otherwise, return to step (A6). The second threshold is set according to actual needs. To avoid endless calculations without outputting a specific compensation value, Figure 4-1 In the embodiment shown, the second threshold is set to 20.

[0124] (A15) End. At this point, one round of calculation is complete, and the basic values ​​for angle compensation and torque offset compensation are obtained.

[0125] like Figure 4-2 As shown, the second step of calculating the mean LKA request angle and the mean C0 includes:

[0126] (B1) Settings: Calculation count x = 0, Monitoring count y = 0;

[0127] (B2) Determine whether the condition is met. If yes, proceed to step (B3); otherwise, proceed to step (B11).

[0128] (B3) Calculate the mean over 30 seconds and assign it to N(x); the vehicle itself samples at regular intervals (e.g., 0.1 seconds). The mean over 30 seconds is obtained by summing the sampled data over 30 seconds and dividing by the number of samples. For example, to find the mean of C0, sum all C0 samples over 30 seconds and divide by the number of samples. Similarly, the mean of the LKA requested angle over 30 seconds can be calculated. For the sake of simplicity, in Figure 4-2 The text only states "the average value over 30 seconds".

[0129] (B4) x = x + 1;

[0130] (B5) Determine whether x < the first threshold is true. If yes, return to step (B2). If no, proceed to step (B6).

[0131] (B6) Determine whether x = 1 is true. If yes, proceed to step (B7); otherwise, proceed to step (B8).

[0132] (B7) Mean = N(x-1), then proceed to step (B13); Proceeding to step (B7) indicates that only the mean within 30 seconds has been calculated once, that is, only N(0). At this time, N(0) can be assigned to the mean. Here, "mean" refers to the C0 mean and the LKA requested angle mean.

[0133] (B8) Determine whether x = 0 is true. If yes, proceed to step (B9); otherwise, proceed to step (B10).

[0134] (B9) Mean = 0, then proceed to step (B13); entering this step indicates that the condition has not been met, therefore no calculation has been performed. Here, "mean" refers to the C0 mean and the LKA requested angle mean, which are set to 0 in this step.

[0135] (B10) Mean = 0.5N(x-1) + 0.5N(x-2), then proceed to step (B13); at this point, take the average of the last two calculated means as the mean, where "mean" refers to the C0 mean and the LKA requested angle mean;

[0136] (B11), y = y + 1;

[0137] (B12) Determine whether y < the second threshold is true. If yes, return to step (B2); otherwise, return to step (B6).

[0138] (B13) End. At this point, one round of calculation is complete, yielding the mean C0 value and the mean LKA requested angle.

[0139] When the vehicle is traveling straight, the steering wheel angle is considered to be at the zero position. To avoid overcorrection caused by the adaptive compensation mechanism, the triggering conditions need to be restricted. The operation of determining whether the conditions are met in steps (A2) and (B2) specifically includes:

[0140] If all four of the following conditions are met simultaneously and last for 30 seconds, the conditions are considered met; otherwise, the conditions are considered not met:

[0141] Condition 1: |C1| < 0.02;

[0142] Condition 2: |C2| < 0.00003;

[0143] Condition 3: V > 10 kph;

[0144] Condition 4: A 1-second delay after the steering wheel torque is less than 0.5 Nm.

[0145] Figure 4-1 and Figure 4-2 The first few steps in the flowchart shown are basically the same. In actual processing, the calculation of four averages is also performed simultaneously. That is, when the conditions are met, the average value of the steering wheel angle, the average value of the LKA angle request, the average value of C0, and the average value of the steering wheel zero-position torque are calculated simultaneously. Then, the average values ​​of the steering wheel angle and the average value of the steering wheel zero-position torque are further processed using... Figure 4-1 The subsequent steps process the data to obtain the baseline values ​​for angle compensation and torque offset compensation. The average values ​​of the LKA requested angle and C0 are then further processed using... Figure 4-2 The subsequent steps process the data to obtain the mean LKA requested angle and the mean C0. If the conditions are never met, meaning no calculation has been performed, all values ​​are set to 0. For ease of description, the overall process of one round of calculation is broken down as follows: Figure 4-1 and Figure 4-2 Two processes.

[0146] Furthermore, if a driver attempts to change lanes, overtake, or avoid obstacles by taking over the steering wheel, it can also cause a significant discrepancy between the actual current angle and the system's real-time planned angle. For example, if the intelligent driving system plans an angle of 0°, but the driver, for some reason, needs to manually turn the steering wheel to 10°, this will result in a 10° error between the system's real-time planned angle and the actual angle. Figure 6 As shown, this creates a counterforce between the steering wheel controlled by the intelligent driving system and the driver. Therefore, the intelligent driving system needs to monitor the driver's takeover behavior more meticulously in real time to eliminate this resistance.

[0147] To reduce the angle error caused by the above-mentioned takeover process, this invention adds an attenuation coefficient α. The requested angle is adjusted in real time using the attenuation coefficient α, making the requested angle closer to the actual steering wheel angle, thereby reducing control error and lowering the output torque of the gear angle controller. This improves the resistance torque felt by the driver. This invention achieves damping compensation through the attenuation coefficient α, which improves the feel when the driver takes over LKA (Lane Kinematics Control). It addresses the problems of stiff, abrupt, and difficult-to-intervene steering under conventional angle control, or the resistance requiring significant force to take over. Furthermore, after the driver takes over, the control effect is controllable, avoiding overshoot and oscillation issues.

[0148] Specifically, the operation of calculating the attenuation coefficient in the second step includes:

[0149] The attenuation coefficient α is calculated using the following formula:

[0150] α = (1-β) × 100%

[0151] β=1, when T C ≤T min hour

[0152] When T min <T C <T max hour

[0153] β=0, when T C ≥T max hour

[0154] Where Tc is the current steering wheel torque, T min The lower limit of steering wheel torque (a value already set in the system), T max β represents the upper limit of steering wheel torque (a value already set in the system), and β is the torque process variable. When the driver manually turns the steering wheel, the intelligent driving system will monitor the steering wheel torque and adjust it by setting a damping coefficient. c Less than or equal to the lower limit of steering wheel torque T min At that time, the angle of the intelligent driving system does not decrease at all; when T c Greater than or equal to the upper limit of steering wheel torque T max At that time, the angle of the intelligent driving system completely decays, when T c When the value is between two values, linear interpolation is used for attenuation.

[0155] The third step is to determine whether compensation is needed. If so, determine the compensation direction, then add the compensation direction before the angle compensation base value to obtain the angle compensation value, and add the compensation direction before the torque offset compensation base value to obtain the torque offset compensation value, then proceed to the fourth step; if not, return to the second step (via...). Figure 4-1 and Figure 4-2 The process involves continuously judging whether the conditions are met using real-time C1, C2, V, and steering wheel torque. When the number of calculations reaches the first threshold or the number of monitoring reaches the second threshold, the current round of calculation is completed. If it is determined that no compensation should be made, the process returns to the second step to start the next round of calculation.

[0156] After obtaining the baseline values ​​for angle compensation, torque offset compensation, LKA requested angle, C0, and attenuation coefficient, it is also necessary to determine whether compensation should be performed and the sign of the compensation value. Figure 5-1 and Figure 5-2 This is a flowchart of the decision module. For example... Figure 5-1 As shown, the operation of determining whether to perform compensation in the third step includes:

[0157] (S1) Determine whether the average value of the LKA requested angle and the average value of C0 are both equal to 0. If yes, proceed to step (S3); otherwise, proceed to step (S2).

[0158] (S2) Determine whether the directions of the LKA requested angle mean and C0 mean are opposite (if the two values ​​are positive and negative, it proves that the vehicle control is not centered or the control value range is insufficient, so compensation is required). If yes, determine to perform compensation and then proceed to step (S4). If no, proceed to step (S3).

[0159] (S3), it is determined that no compensation will be provided; from Figure 4-2 As can be seen from the process, the mean of the LKA requested angle and the mean of C0 will only equal 0 when no calculations have been performed. Figure 4-1 As can be seen from the process, the base compensation value is also 0 at this time. Therefore, no compensation is required in this case.

[0160] (S4), End.

[0161] like Figure 5-2 As shown, the operation of determining the compensation direction in the third step includes:

[0162] Determine if the mean C0 value is greater than a threshold. If it is, the compensation direction is determined to be negative; otherwise, the compensation direction is determined to be positive. Generally, the driver's side is used, with left being negative and right being positive. The threshold can be set based on practical experience.

[0163] The fourth step is to obtain the adaptively compensated planned angle based on the angle compensation value and the attenuation coefficient, and then send the adaptively compensated planned angle and the torque offset compensation value to the EPS.

[0164] The operation in the fourth step of obtaining the adaptively compensated planning angle based on the angle compensation value and the attenuation coefficient includes:

[0165] The adaptively compensated planning angle θt is calculated using the following formula:

[0166] θt=θc+α(θp-θc)K+θa,

[0167] Where: θc is the current steering wheel angle, θp is the planned angle, θa is the angle compensation value, and K = calibration constant coefficient (the value of K is an empirical parameter).

[0168] In the fifth step, the EPS converts the adaptively compensated planned angle into a planned torque. Then, it adds the planned torque to the torque offset compensation value (the torque offset compensation value has a positive or negative sign; simply add the planned torque and the torque offset compensation value directly) to obtain the total executed torque, and then returns to the second step. The EPS motor controls the vehicle based on the total executed torque. After this round of adaptive compensation is completed, it returns to the second step to start a new round of calculation, thus continuously performing adaptive compensation in real time based on the real-time input parameters.

[0169] Currently, after receiving angle commands from the intelligent driving controller, the EPS (Electric Power Assist) uses the current vehicle speed, the torque applied to the steering wheel by the driver (i.e., the driver's hand force), and the current steering wheel angle to look up the basic assist curve, damping torque curve, and centering torque curve, and then calculates and outputs the steering wheel torque. Meanwhile, the LKA (Lane Assist Key) responds to the command value in the shortest possible time to ensure good control performance. However, during the function activation phase, if the requested angle from the controller deviates too much from the actual current angle, it will cause the steering wheel to rotate rapidly, making the driver feel a large force on the steering wheel. This invention solves the influence of zero-point torque in the EPS through torque offset compensation and angle compensation, improving the control accuracy of the EPS, resolving the issues of LKA control not being centered and exhibiting a serpentine pattern, and improving the LKA control accuracy when the vehicle chassis is loose and the vehicle is veering. Simultaneously, damping compensation provides comfort control for the EPS, ensuring that the angle response and disengagement are not too aggressive, but rather a gradual rise and fall process, improving the driver's comfort during control and thus optimizing hand force.

[0170] The above technical solution is only one embodiment of the present invention. For those skilled in the art, based on the principles disclosed in the present invention, it is easy to make various types of improvements or modifications, and not limited to the technical solutions described in the specific embodiments of the present invention. Therefore, the foregoing description is only preferred and not restrictive.

Claims

1. An adaptive compensation lane keeping system, characterized by: The system comprises an input module, a calculation module, a judgment module and an output module connected in sequence; The input module is used for inputting parameters; The calculation module is used for calculating an angle compensation base value, a torque bias compensation base value, an LKA request angle average value, a C0 average value and a decay coefficient; The judgment module is used for judging whether compensation is needed according to the LKA request angle average value and the C0 average value, and if so, determining a compensation direction, and adding the compensation direction in front of the angle compensation base value to obtain an angle compensation value, and adding the compensation direction in front of the torque bias compensation base value to obtain a torque bias compensation value; The output module is used for obtaining a planning angle after adaptive compensation according to the angle compensation value and the decay coefficient, and sending the planning angle after adaptive compensation and the torque bias compensation value to an EPS; The C0 refers to a relative distance between a driving estimation line of a current vehicle position and a vehicle body.

2. The self-compensating lane keeping system of claim 1, wherein: The parameters inputted by the input module include: a vehicle speed V, a yaw rate, a relative distance C0 between a driving estimation line of a current vehicle position and a vehicle body, a heading angle C1, a target trajectory curvature C2 of a current vehicle position, a curvature change rate C3 of a target trajectory, a steering wheel torque, an LKA control state, a steering wheel angle, an LKA request angle.

3. A method of adaptive compensation for lane keeping, characterized by: The method comprises: a first step of inputting parameters: a vehicle speed V, a yaw rate, a relative distance C0 between a driving estimation line of a current vehicle position and a vehicle body, a heading angle C1, a target trajectory curvature C2 of a current vehicle position, a curvature change rate C3 of a target trajectory, a steering wheel torque, an LKA control state, a steering wheel angle, an LKA request angle; a second step of calculating an angle compensation base value, a torque bias compensation base value, an LKA request angle average value, a C0 average value and a decay coefficient; a third step of judging whether compensation is needed, and if so, determining a compensation direction, then adding the compensation direction in front of the angle compensation base value to obtain an angle compensation value, and adding the compensation direction in front of the torque bias compensation base value to obtain a torque bias compensation value, and then entering a fourth step; if not, returning to the second step; a fourth step of obtaining a planning angle after adaptive compensation according to the angle compensation value and the decay coefficient, and sending the planning angle after adaptive compensation and the torque bias compensation value to an EPS; a fifth step of converting the planning angle after adaptive compensation into a planning torque by the EPS, then superimposing the planning torque and the torque bias compensation value to obtain a total execution torque, and then returning to the second step.

4. The method of adaptive compensation for lane keeping as claimed in claim 3, characterized in that: The operation of calculating the angle compensation base value and the torque bias compensation base value in the second step comprises: (A1), setting: a calculation number x = 0, a monitoring number y = 0, a memory value = 0; (A2), judging whether a condition is met, if so, entering step (A3), if not, entering step (A13); (A3), calculating an average value within 30 seconds and assigning it to N(x); (A4), x = x + 1; (A5), judging whether x < a first threshold value is true, if so, returning to step (A2), if not, entering step (A6); (A6), judging whether x = 1 is true, if so, entering step (A7), if not, entering step (A11); (A7), y = 0, and entering step (A8); (A7) mean = N(x-1); (A8) judge whether the absolute value of the difference between the mean value and the memory value is greater than a set threshold value, if yes, go to step (A9), if no, go to step (A10); (A9) compensation base value = mean value, memory value = compensation base value; then go to step (A15); (A10) compensation base value = 0, memory value = 0; then go to step (A15); (A11) judge whether x = 0 is true, if yes, return to step (A10), if no, go to step (A12); (A12) mean = 0.5N(x-1) + 0.5N(x-2); then return to step (A8); (A13) y = y + 1; (A14) judge whether y < second threshold value is true, if yes, return to step (A2), if no, return to step (A6); (A15) end; When calculating the angle compensation base value, the compensation base value is the angle compensation base value, and when calculating the torque bias compensation base value, the compensation base value is the torque bias compensation base value.

5. The method of adaptive compensation for lane keeping as claimed in claim 4, characterized in that: The operation of calculating the LKA request angle mean value and the C0 mean value in the second step comprises: (B1) setting: calculation times x = 0, monitoring times y = 0; (B2) judge whether the condition is met, if yes, go to step (B3), if no, go to step (B11); (B3) calculate the mean value within 30 seconds and assign it to N(x); (B4) x = x + 1; (B5) judge whether x < first threshold value is true, if yes, return to step (B2), if no, go to step (B6); (B6) judge whether x = 1 is true, if yes, go to step (B7), if no, go to step (B8); (B7) mean = N(x-1); then go to step (B13); (B8) judge whether x = 0 is true, if yes, go to step (B9), if no, go to step (B10); (B9) mean = 0; then go to step (B13); (B10) mean = 0.5N(x-1) + 0.5N(x-2); then go to step (B13); (B11) y = y + 1; (B12) judge whether y < second threshold value is true, if yes, return to step (B2), if no, return to step (B6); (B13) end; When calculating the C0 mean value, the mean value is the C0 mean value, and when calculating the LKA request angle mean value, the mean value is the LKA request angle mean value.

6. The method of adaptive compensation for lane keeping as claimed in claim 5, characterized in that: The operation of judging whether the condition is met in the steps (A3) and (B3) comprises: If the following four conditions are met at the same time and last for 30 seconds, it is determined that the condition is met, otherwise it is determined that the condition is not met: Condition one: |C1| < 0.02; Condition two: |C2| < 0.00003; Condition three: V > 10 kph; Condition four: steering wheel torque < 0.5 N.m after 1 s delay.

7. The method of adaptive compensation for lane keeping as claimed in claim 3, wherein: The operation of calculating the attenuation coefficient in the second step comprises: The attenuation coefficient α is calculated by using the following formula: α＝(1-β)×100％； β = 1 when T C ≤ T min ; When T min <T C <T max time; β = 0 when T C ≥ T max ; where T c is the current steering wheel torque, T min is the lower steering wheel torque limit, T max is the upper steering wheel torque limit, and β is the torque process variable.

8. The method of adaptive compensation for lane keeping as claimed in claim 3, wherein: The operation of determining whether to compensate in the third step comprises: (S1), determining whether the LKA request angle average and the C0 average are equal to 0, if yes, entering step (S3), if no, entering step (S2); (S2), determining whether the directions of the LKA request angle average and the C0 average are opposite, if yes, determining to compensate, and then entering step (S4), if no, entering step (S3); (S3), determining not to compensate; (S4), ending.

9. The method of adaptive compensation for lane keeping as claimed in claim 3, wherein: The operation of determining the compensation direction in the third step comprises: determining whether the C0 average is greater than a threshold, if yes, determining the compensation direction to be negative, if no, determining the compensation direction to be positive.

10. The method of adaptive compensation for lane keeping as claimed in claim 7, wherein: The operation of obtaining the planning angle after adaptive compensation according to the angle compensation value and the attenuation coefficient in the fourth step comprises: obtaining the planning angle after adaptive compensation θt by calculation using the following formula: θt = θc + α(θp - θc)K + θa, wherein θc is the current steering wheel angle, θp is the planning angle, θa is the angle compensation value, and K = calibration constant.

Images